https://bugzilla.kernel.org/show_bug.cgi?id=209177 --- Comment #3 from sun.nagarajan@xxxxxxxxx --- (2020-09-30): Suganath Prabu Subramani suggested the following: >From log, I could see that the HBA queue depth is very high "32455" as shown below. [ 11.465416] mpt2sas_cm0: hba queue depth(32455), max chains per io(128). In this patch "https://patchwork.kernel.org/patch/11505139/"; driver is allocating the DMA-able memory for RDPQ's in sets of 16 reply queues using limitation of Ventura series controller. With 32455 queue depth and above patch, Driver may request a large DMA-able memory where the kernel may fail to allocate. To confirm this, Please try by tuning the queue depth to 8000/10000 using the module parameter "mpt3sas.max_queue_depth=10000". With the same unmodified 5.8.12 kernel, boot completed with no errors. Attaching dmesg.mpt3sas.max_queue_depth.20200930 -- You are receiving this mail because: You are on the CC list for the bug.
